LAISHA
Open Laisha

GATE ST Linear Models & Regression Analysis: practice questions

GATE - Statistics (ST) · India · IITs and IISc on rotation (2027: IIT Madras)

Original practice questions written to the published syllabus · not past papers
10worked here, free
28written to this exam in this subject
8syllabus units

About GATE ST

GATE - Statistics (ST) is a postgraduate admission exam run by IITs and IISc on rotation (2027: IIT Madras), for candidates in India. Computer-based, 3 hours, 100 marks (15 General Aptitude + 85 subject). MCQ, MSQ and NAT; negative marking on MCQ only; on-screen virtual calculator.

Source: gate2026.iitg.ac.in.

Checked against the exam body's published information. Formats and fees change; confirm the current edition there before you book.

Linear Models & Regression Analysis in GATE ST

The general linear model in matrix form: estimability, the Gauss-Markov theorem, least squares with full and less-than-full rank, the analysis of variance as a linear model, model checking and selection, and regression diagnostics. It is the theory under every design in Phase 3.

Our 28 GATE ST questions in this subject: 7 easy, 17 medium, 4 hard, across 8 units.

10 practice questions, worked

Choose an answer, then open the working. Each explanation says why the right answer is right and why the others are not.

These 10 were written to GATE ST and are not shown on any other exam's page.

Question 1 · medium

What is the variance matrix of the least squares estimator under V(e) = sigma^2 I?

  1. sigma^2 X(X'X)^(-1)X'
  2. sigma^2 (XX')^(-1)
  3. sigma^2 (X'X)^(-1)
  4. sigma^2 X'X
Answer and working

C. sigma^2 (X'X)^(-1)

V((X'X)^(-1)X'y) = (X'X)^(-1)X' (sigma^2 I) X(X'X)^(-1) = sigma^2 (X'X)^(-1). sigma^2 H is the variance of the fitted values.

Question 2 · hard

In simple linear regression, the variance of the fitted mean response at x0 is smallest when:

  1. x0 is zero, where the intercept is estimated
  2. x0 is at the smallest observed x value
  3. x0 equals the mean of the x values
  4. x0 is at the largest observed x value
Answer and working

C. x0 equals the mean of the x values

V = sigma^2 [1/n + (x0 - x-bar)^2/Sxx], minimised at x0 = x-bar; precision falls as x0 moves away, especially outside the data.

Question 3 · medium

If y is N(0, sigma^2 I) and A is symmetric, y'Ay/sigma^2 has a chi-square distribution if and only if:

  1. A is diagonal
  2. A is positive definite
  3. A has determinant one
  4. A is idempotent
Answer and working

D. A is idempotent

The chi-square property holds iff A^2 = A; the degrees of freedom are rank(A) = trace(A).

Question 4 · medium

For n = 10, what are the degrees of freedom of y'(I - J/n)y / sigma^2, the corrected sum of squares over sigma^2, under normality?

  1. 8
  2. 10
  3. 1
  4. 9
Answer and working

D. 9

I - J/n is idempotent with rank n - 1 = 9. J/n has rank 1 and gives the 1-df sum of squares for the mean.

Question 5 · hard

Two quadratic forms y'Ay and y'By in y ~ N(mu, sigma^2 I) are independent if and only if:

  1. AB = 0
  2. A = B'
  3. A + B = I
  4. AB = BA
Answer and working

A. AB = 0

Independence of quadratic forms in normal vectors requires AB = 0 (Craig's theorem). Commuting matrices are not enough.

Question 6 · hard

Cochran's theorem states that if y'y is split into quadratic forms y'A_i y with y ~ N(0, sigma^2 I), they are independent chi-squares if and only if:

  1. Each A_i has rank one
  2. The traces of the A_i are all equal
  3. Each A_i is diagonal
  4. The ranks of the A_i add up to n
Answer and working

D. The ranks of the A_i add up to n

Sum of ranks = n is the necessary and sufficient condition; this is why ANOVA degrees of freedom must add up.

Question 7 · easy

A multiple regression with k = 4 predictors and n = 25 observations has R^2 = 0.8. What is the adjusted R^2?

  1. 0.960
  2. 0.833
  3. 0.760
  4. 0.762
Answer and working

C. 0.760

Adjusted R^2 = 1 - (1 - R^2)(n - 1)/(n - k - 1) = 1 - 0.20 x 24/20 = 0.760. Inverting the ratio gives 0.833, larger than R^2, which is impossible.

Question 8 · easy

What happens to R^2 when an extra predictor is added to a regression model?

  1. It never decreases
  2. It always decreases
  3. It always stays exactly the same
  4. It decreases if the predictor is useless
Answer and working

A. It never decreases

RSS cannot increase when a column is added, so R^2 cannot fall; adjusted R^2 can fall.

Question 9 · easy

In a regression with p = 4 parameters and n = 40 observations, what is the average leverage?

  1. 0.111
  2. 0.20
  3. 0.10
  4. 0.025
Answer and working

C. 0.10

The leverages sum to trace(H) = p, so the average is p/n = 4/40 = 0.10; the common flag is 2p/n = 0.20.

Question 10 · medium

What is the ridge regression estimator?

  1. (X'X - kI)^(-1) X'y with k > 0
  2. (X'X + kI)^(-1) X'y with k > 0
  3. k (X'X)^(-1) X'y with 0 < k < 1
  4. (X'X)^(-1) X'y + kI
Answer and working

B. (X'X + kI)^(-1) X'y with k > 0

Adding k to the diagonal of X'X stabilises the inverse and shrinks coefficients; k = 0 returns OLS.

A study plan for Linear Models & Regression Analysis in GATE ST

This plan follows where our questions actually fall, so the time goes where the marks are likely to be. It is our reading of the syllabus, not the exam body's weighting.

  1. First, the heavy units (71% of the 28 questions): The general linear model and least squares in matrix form (9, 32%); Testing linear hypotheses; R-squared and adjusted R-squared (6, 21%); Distribution of quadratic forms and the Fisher-Cochran theorem (5, 18%).

    Read the manual chapters for these first and do every worked question you can find on them.

  2. Then the middle band: Estimable functions and the Gauss-Markov theorem (2, 7%); Examination of residuals and model adequacy (2, 7%); Influence and leverage: Cook's distance and hat values (2, 7%).

    Aim for one sitting per unit, then mixed questions across them.

  3. Then the lighter units: Multicollinearity and ridge regression (1, 4%); Weighted and generalised least squares (1, 4%).

    Learn the definitions, classifications and standard numbers; these units are more often recalled than worked.

  4. Leave time for the hard ones: Distribution of quadratic forms and the Fisher-Cochran theorem (2 hard of 5), Testing linear hypotheses; R-squared and adjusted R-squared (1 hard of 6), The general linear model and least squares in matrix form (1 hard of 9).

    Come back to these after a first pass, untimed, and read every explanation, including for the options you rejected.

  5. Bank the easy marks: Influence and leverage: Cook's distance and hat values (1 easy of 2); The general linear model and least squares in matrix form (4 easy of 9).

    Do these early in revision so they are fast on the day.

  6. Finish timed. Sit the subject as a timed set without notes, mark it, and send each wrong answer back to its unit above.

Last updated · How Laisha writes and checks its pages · Report a mistake

Ask Laisha about this

Opens Laisha's chat with your question ready to send. Sign in or start free.

Unlock every question — GATE ST Linear Models & Regression Analysis

US$10 a month · one price worldwide · cancel any time

  • Every practice question in this subject, with full working
  • Timed exam papers and progress by unit
  • Ask Laisha to explain any answer — up to 1,000 questions a month
Free or Smarter
FreeSmarter
Grow guides, pest pages, articlesRead allRead all
Ask Laisha5 a dayUp to 1,000 a month
Exam practiceThe sample on each pageEvery question, timed papers
Crop calendar and pest alerts—For your own fields
Calculators and tools—All
Uploads and reports—100 a month, real files
10 days free for every new account, no card.